RE: Math Working Group Charter

The MATH table kerning is for script-base pairs, such as in V_c. The subscript c is pulled in under the V a bit. Accents are placed using the OpenType gpos table.

Murray

From: Paul Topping [mailto:pault@dessci.com]
Sent: Tuesday, February 18, 2014 2:24 PM
To: Frederic Wang; www-math@w3.org
Subject: RE: Math Working Group Charter

I believe that the MATH table only contains kerning info for math accents and the like. The kerning info for normal text must lie in another table as it predates the invention of the MATH table.

It has been noted elsewhere that the so-called Web Platform is becoming the equivalent of the modern OS. It should eventually have all the facilities that modern Oss have. At least the ones that make sense. Both Windows and Mac Oss allow access to kerning info and even the raw data in the OpenType tables.

Since you mention that Gecko will need access to kerning info to do its job properly, isn’t that also true of MathJax by extension? Ideally, MathJax (and any other JS code in the browser) should be able to work with fonts much like the browser implementation. This is not to say that many apps will need it, of course. Most will use higher-level facilities.

Paul

From: Frédéric WANG [mailto:fred.wang@free.fr]
Sent: Tuesday, February 18, 2014 2:03 PM
To: www-math@w3.org<mailto:www-math@w3.org>
Subject: Re: Math Working Group Charter

Le 18/02/2014 22:25, Paul Topping a écrit :
On the web fonts, I really didn’t have an agenda. I just know that math always has special requirements and there might be some areas where we need some help. Recently, a question came up about access to kerning info. Do webfonts currently allow JS code to access such things? Same with the MATH table. Although the webfonts implementation should make use of this data, might not JS code benefit from access to it as well?
The kerning info is given in the MATH table so it will be accessible to browsers, but they will need to use it for the MathML rendering (the Gecko  bug is https://bugzilla.mozilla.org/show_bug.cgi?id=961482 but I have not started to work on that yet ; no plan for WebKit for now since there are more important issues to fix first).

The JS code can not have access to the kerning info but it does not have access to the MATH table either and even don't have access to basic font metrics that are unrelated to math nor to non-Unicode glyphs. In my opinion, that's a bit overkill to create JS API to expose all the font data just for the purpose of MathML polyfills especially if we see them as workaround before good native MathML support. I believe individuals and companies interested in this MATH feature as well as in compatibility with the Web platforms in general should focus on native MathML and help browser developers to create clean and efficient implementation rather than spending time on standardizing temporary hacks.

Also, for polyfills, one can just extract the kerning info from Open Type MATH table via a Python script and generate JS files with this info. This is what MathJax does for font metrics and for size variants and components of stretchy operators (https://github.com/mathjax/MathJax-dev/tree/master/fonts/OpenTypeMath) and this can be done for all the MATH table parameters.

(note: the distinction between web or system fonts is irrelevant when talking about which info available to browsers or Javascript so that's why I didn't see why you mentioned Web fonts specifically)



--

Frédéric Wang

MathML Crowdfunding: ulule.com/mathematics-ebooks